This invention concerns a heterostructure semiconductor laser for producing much higher power levels of coherent radiation than heretofore available. By having distributed feedback in the laser, a single frequency without significant side bands is produced.
The power output of GaAlAs laser diodes have generally been limited to output powers of less than 50 milliwatts. There are various reasons for such limitations. For example, when the power density at the output facet of the laser is too high, there can be catastrophic optical mirror damage. This is believed to be due to the intense absorption of the laser radiation at the interface between the active layer and the air. Because of carrier depletion caused by highly efficient carrier recombination at the surface, the active layer material becomes absorptive, and the resulting temperature rise causes a localized drop in the semiconductive material band gap and increased absorption. The ensuing thermal runaway ends in sudden melting or spalling of the active layer at the facet.
The power density at which catastrophic mirror damage occurs can be increased by providing non-absorbing "windows" at the end of the active layer. A thin layer of material is provided between the active gain medium and the end facet. This material has a band gap larger than the energy of the laser radiation, and hence does not result in energy absorption.
Thermal saturation can also limit power output from a semiconductor laser. Temperature of the active layer may increase due to non-radiative carrier recombination, by absorption of both spontaneous and stimulated radiation, and by ohmic heating. As the temperature of the active layer rises, its gain coefficient falls, resulting in lower stimulated output for given injected current than would be the case at lower temperatures. A point is eventually reached for which an increase in current results in no increase in output power. There are ways of limiting thermal saturation by providing heat sinks that effectively withdraw heat from the active layer and by careful design to minimize ohmic heating. However, there are limits to what can be done for increasing power.
High power levels can also result in abrupt changes in the dominant transverse mode in the laser, resulting in "steering" of the output beam or appearance of side lobes in the far field pattern.
Various techniques have been employed for enhancing the power levels of semiconductor lasers. The non-absorbing facets to avoid catastrophic mirror damage have increased power outputs be an order of magnitude. Power densities of 25 megawatts per square centimeter have been obtained at the output facet. Continuous wave lasing of buried heterostructure devices only 1.2 micrometers wide have given power levels as high as 175 milliwatts. Similar devices without non-absorbing facets fail at only 10 milliwatts.
Alternatively, since mirror damage is caused by high power densities, if the power is spread over a large area of the facets, greater amounts of total power can be obtained. Simply widening the pumped region of the laser to hundreds of microns has proved unsuccessful because, instead of the entire pumped region lasing coherently, bright localized "filaments" form at random locations across the output facet, and their position and intensities vary widely with current. The resulting output of most broad area lasers is poorly coherent and not diffraction limited. Reproducibility of results are generally poor.
A variation of this approach is to try to phase lock arrays of lasers in the so-called "fundamental super mode" resulting in a single lobed far field. Although significant amounts of raw power can be obtained, the arrays show a strong tendency to double lobed output where separate laser stripes are out of phase with each other. Occasional satisfactory results are obtained, but reproducibility is low. Reliability tends to be low since there are several failure modes which can prevent arrays from performing satisfactorily.
Still another approach is to provide a relatively low power laser and optically couple it to an active optical amplifier. The quality of the output from the laser can be controlled since it is operated at power levels where high quality, coherent, diffraction limited output can be obtained. The beam is then passed through an active gain medium which does not have substantial oscillation, to amplify the power without severely degrading the beam. Some beam degradation is essentially inevitable. Such optical amplifiers can produce relatively high power levels since the power density at the output can be maintained at tolerable levels to prevent catastrophic mirror damage.
One type of laser amplifier which has been proposed comprises a diverging active gain layer similar to the active layer of the laser and having an input facet aligned with the output facet of the laser. The beam spreads by diffraction in the amplifier and the power density at the output facet of the amplifier is tolerable. Such a technique presents difficult fabrication problems since precise alignment is required between the laser output and amplifier input. Tolerances are in the order of a fraction of a micrometer.
Conventional semiconductor laser cavities do not emit radiation in a single longitudinal mode. Typically, there is a main frequency peak where much, if not most, of the radiation energy is concentrated. A few percent of the output power is carried by other longitudinal modes lying close to the gain peak. These decrease the power in the center frequency, particularly at high modulation rates. Further, the unwanted side modes limit the information transmission rate. A single stable longitudinal mode is desirable for communication where heterodyne demodulation is employed for enhancing the signal to noise ratio. A stable longitudinal mode is desirable for both the transmitter of radiation an for the local oscillator at the receiver.
It is, therefore, desirable to devise means for a semiconductor laser to emit radiation predominantly in a single longitudinal mode, even under high speed modulation. Such lasers are referred to as single-frequency lasers.
It is therefore desirable to provide a single high power laser which is not limited by catastrophic mirror damage at its output facet, which does not require precision manufacturing techniques beyond the current state of the art for semiconductor lasers, which provides a high quality, coherent, diffraction limited output beam, which has a single output frequency, which does not require precision control to avoid extraneous output modes, and which can be modulated at high frequency.